Microscopic pathology defines many human diseases, but the chemical composition of most histologically observed subcellular structures remains largely uncharacterized. Currently, there is no method for isolating subcellular areas of interest without disrupting the cells and thus distorting naturally occurring protein interactions. We propose to combine in situ labeling schemes and subsequent microdissection with mass spectrometry to directly analyze subcellular complexes and organelles. To this end, we are improving expression microdissection (xMD) in order to isolate optically dark targets (chemical, antibody, or metal-based stains) using optimized protocols compatible with mass spectrometric analysis to allow unsupervised thermoplastic capture of subcellular structures. In contrast to Laser Capture Microdissection, the film in xMD does not contain a dye and consequently the heat required for thermoplastic melting and adhesion is derived from the stained organelle. xMD is compatible with microscopic methodologies and avoids tissue homogenization. Combining the xMD technique with subsequent liquid chromatography coupled nano-spray mass spectrometry (LC-MS/MS) provides a novel and unique analytical method to identify the composition of subcellular components and characterize their physiological functions in cells. For xMD, a clear composite thin film consisting of 0.5-2 microns of a hot-melt adhesive (EVA) on a flexible substrate (12.5 micron thick PET) is placed on the top of immunohistochemically stained formalin-fixed paraffin- embedded or frozen rat brain tissue(5-10 micron thick). The EVA layer was made by spin-coating from toluene solutions with varying percentages (4 to 10% w/v) of EVA while the flexible backing film was temporarily held in place on a rigid glass substrate using capillary forces. The heat generated by optical absorption of a full spectrum xenon lamp over the slide melts the film in stained areas and provides selective adhesion to the targets. When the film is lifted, the targeted stained organelles are captured for subsequent mass spectrometric analysis. Significantly, the resolution for capture is not dependent on the optical resolution of the system, but is instead determined by the spatial extent of the stain and the thermal properties of the irradiated assembly. An in-house prototype xMD instrument was used to dissect 60-80 % of neuronal nuclei from formalin fixed paraffin embedded rat brain tissue (10 micron). Equal success was achieved with either histone-immunohisto- chemical- or hematoxylin-chemical stained tissues. The selectivity of transfer of nuclei onto the EVA film was dependent upon multiple controllable variables, but consistent results demonstrated that the xMD method is feasible for isolation of subcellular organelles in an unsupervised mode. Scanning electron microscopic images of the transferred nuclei suggest that there are accompanying fragments of endoplasmic reticulum transferred with some of the nuclei. We tested multiple solvent solubilization and digestion methods for the analysis of DAB-peroxidase immunohistochemically stained proteins. A trifluoroethanol-ammonium bicarbonate on-film trypsin digestion protocol was determined to be most successful for microdissected samples. The resulting tryptic peptides were analyzed using LC-MS/MS on nanospray-high resolution orbitrap instruments. MS/MS spectra were matched to peptide sequences using automated searching (Mascot), and then a parsimonious list of proteins in nuclei assigned using MassSieve and manual inspection. For comparison purposes, nuclei were isolated from fresh rat brain tissue using conventional methods (sucrose density gradient protocol). Results from measurable concentrations of fresh, unstained nuclei indicate that the amount of protein captured on-film corresponds to approximately 150 nanograms per sq. cm. From that quantity, 350 proteins could be identified with two or more peptides in the conventionally isolated nuclei, and 288 proteins from the hemotoxylin stained, formalin fixed paraffin embedded tissue, corresponding to an overlap of 229 proteins (80% of the xMD set). The xMD isolated peptides were 20% enriched in arginine vs. lysine C-termini, probably a result of blocked lysines due to formylation during fixation. The absence of certain characteristic nuclear proteins in the xMD sample may be the result of the same phenomenon. Gene ontology compartment analysis shows that the xMD method yielded a comparable percentage of nuclear proteins as the conventional fractionation method. Quantitative measures of proteins characteristic of nuclei vs. other organelles are planned. In summary, immunohistochemical and chemical guided unsupervised xMD combined with LC-MS/MS has proven to be reproducible and feasible by experiments on DAB- and hemotoxylin stained neuronal nuclei of rat brain, and the specificity of nuclear transfer is good. Multiple capture films can be combined using a modified methanol-chloroform protein precipitation step to improve peptide recovery. The recovery of peptides is sufficient for subsequent mass spectrometric proteomics. Data suggests subcellular organelle enrichment using xMD with subsequent proteomic analysis is achievable.